A system and method implemented on a computing device for analyzing a digital corpus of unstructured interlocutor conversations to discover intents, goals, or both intents and goals of one or more parties to the conversations, by grouping the conversation utterances according to semantic similarity clusters; selecting the best utterance(s) that mostly likely embody a party's stated goal or intent; creates a set of candidate intent names for each cluster based upon each intent utterance in each conversation in each cluster; rates each candidate intent (or goal) for each intent name; and selects the most likely candidate intent (or goal) name for the purposes of subsequent automation of future conversations such as, but not limited to, automated electronic responses using Artificial Intelligence and machine learning.
Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
2. The method of claim 1 wherein the grouping is preceded by encoding sentence embeddings contained in the corpus.
The invention relates to natural language processing (NLP) and text analysis, specifically methods for grouping or clustering text data. The problem addressed is efficiently organizing large volumes of text into meaningful groups based on semantic similarity, which is challenging due to the high dimensionality and variability of natural language data. The method involves encoding sentences from a text corpus into numerical embeddings, which are dense vector representations capturing semantic meaning. These embeddings are then used to group similar sentences together. The encoding step ensures that sentences with similar meanings are positioned close to each other in the embedding space, enabling effective clustering. The grouping process may involve techniques such as hierarchical clustering, k-means, or other unsupervised learning methods to form coherent clusters of semantically related sentences. By first encoding sentences into embeddings, the method improves the accuracy and efficiency of text grouping compared to traditional approaches that rely on raw text or simpler feature representations. This is particularly useful in applications like document summarization, topic modeling, and information retrieval, where organizing text data by semantic similarity is critical. The method can be applied to any text corpus, including documents, articles, or user-generated content, to enhance searchability and analysis.
3. The method of claim 2 wherein the encoding sentence embeddings comprises performing Language-Agnostic Bidirectional Encoder Representations from Transformers Sentence Encoding (LABSE).
4. The method of claim 2 wherein the encoding sentence embeddings comprises performing Robustly Optimized Bidirectional Encoder Representations from Transformers Pretraining Approach (RoBERTa).
5. The method of claim 2 wherein the encoding sentence embedding is followed by, prior to the grouping, performing dimensionality reduction on the encoded sentence embeddings.
6. The method of claim 5 wherein dimensionality reduction comprises performing Uniform Manifold Approximation and Projection (UMAP).
7. The method of claim 5 wherein dimensionality reduction comprises performing t-Distributed Stochastic Neighbor Embedding (t-SNE).
8. The method of claim 1 wherein the grouping comprises performing clustering.
A method for organizing data involves grouping data points based on their similarities to improve data analysis and processing efficiency. The grouping process includes performing clustering, which is an unsupervised machine learning technique that partitions data into distinct clusters where data points within the same cluster are more similar to each other than to those in other clusters. This clustering step helps in identifying patterns, reducing dimensionality, and enhancing the accuracy of subsequent data processing tasks. The method may also involve preprocessing the data to remove noise or irrelevant features before clustering, ensuring that the resulting clusters are more meaningful and accurate. By applying clustering, the method enables more efficient data retrieval, classification, and decision-making in various applications such as customer segmentation, anomaly detection, and image recognition. The technique is particularly useful in large-scale datasets where manual analysis is impractical, providing a scalable and automated approach to data organization.
9. The method of claim 8 wherein the clustering comprises performing Kmeans clustering.
10. The method of claim 8 wherein the clustering comprises performing Concensus clustering.
11. The method of claim 1 wherein the selecting is preceded by performing cluster splitting.
12. The method of claim 11 wherein the cluster splitting comprises performing splitting clusters into clusters for label generation and clusters for label ranking.
13. The method of claim 1 wherein the creating of the set of candidate intent names comprises performing label generation.
This invention relates to natural language processing (NLP) and intent recognition systems, specifically addressing the challenge of generating accurate and relevant intent names from user inputs. The method involves creating a set of candidate intent names by performing label generation, which includes analyzing user inputs to identify patterns, keywords, or semantic structures that can be mapped to predefined or dynamically generated intent labels. The system may use machine learning models, such as neural networks or statistical classifiers, to predict the most likely intent names based on the input data. Additionally, the method may incorporate techniques like clustering, topic modeling, or rule-based extraction to refine the candidate intent names. The generated labels are then used to improve the accuracy of intent recognition in applications like chatbots, virtual assistants, or automated customer support systems. The invention aims to enhance the efficiency and precision of intent detection by dynamically adapting to variations in user language and context.
14. The method of claim 13 wherein the label generation comprises performing Generative Pre-trained Transformer 2 (GPT-2).
15. The method of claim 13 wherein the label generation comprises performing Bidirectional Encoder Representations from Transformers (BERT).
The invention relates to natural language processing (NLP) and text analysis, specifically improving the accuracy of label generation for text data. The problem addressed is the inefficiency and inaccuracy of traditional methods in automatically categorizing or labeling text, which often rely on rule-based systems or shallow machine learning models that struggle with context and semantic nuances. The method involves generating labels for text data by leveraging advanced neural network techniques. Specifically, it uses Bidirectional Encoder Representations from Transformers (BERT), a state-of-the-art language model that processes text bidirectionally to capture deep contextual relationships. BERT is trained on a large corpus of text, allowing it to understand complex linguistic patterns, synonyms, and contextual meanings. The method applies BERT to analyze input text, extract meaningful representations, and generate accurate labels based on these representations. This approach enhances the precision of text classification, sentiment analysis, and other NLP tasks by incorporating contextual understanding that traditional methods lack. The system may also include preprocessing steps to clean and normalize text before applying BERT, ensuring optimal performance. The use of BERT significantly improves label accuracy compared to conventional methods, making it suitable for applications requiring high-quality text analysis, such as automated content moderation, customer support, and document organization.
16. The method of claim 13 wherein the creating of the set of candidate intent names comprises performing simplification on the generated labels.
This invention relates to natural language processing (NLP) and intent recognition systems, specifically improving the accuracy and usability of intent classification in conversational AI. The problem addressed is the ambiguity and complexity of intent labels generated from user inputs, which can lead to misclassification or poor user experience. The solution involves refining raw intent labels through a simplification process to produce a more accurate and standardized set of candidate intent names. The method begins by generating initial intent labels from user inputs, typically using machine learning models or rule-based systems. These labels may contain redundant, overly complex, or inconsistent terms. The simplification process involves normalizing the labels by removing unnecessary words, standardizing terminology, and ensuring consistency across similar intents. For example, variations like "book a flight" and "reserve a flight" may be simplified to a common intent name like "flight reservation." The simplified intent names are then used to create a refined set of candidate intents, which are more reliable for downstream classification tasks. This approach enhances the performance of intent recognition systems by reducing ambiguity and improving the clarity of intent definitions. The method is particularly useful in applications like chatbots, virtual assistants, and customer service automation, where accurate intent detection is critical for effective user interaction.
17. The method of claim 1 wherein the step of rating comprises ranking for most likely to least likely using a statistical model trained using a dataset for semantic similarity matching of labels to full sentences.
This invention relates to a method for rating or ranking items, such as labels or full sentences, based on semantic similarity. The method addresses the challenge of accurately matching labels to sentences in natural language processing tasks, where traditional keyword-based approaches often fail to capture contextual meaning. The core innovation involves using a statistical model trained on a dataset specifically designed for semantic similarity matching. This model evaluates the likelihood of a label being semantically similar to a given sentence, then ranks the labels from most likely to least likely based on this evaluation. The statistical model leverages machine learning techniques to improve accuracy in semantic matching, ensuring that the ranking reflects meaningful relationships between labels and sentences. This approach enhances applications in text analysis, information retrieval, and automated labeling systems by providing more precise and context-aware results. The method is particularly useful in scenarios where nuanced language understanding is required, such as in document classification, search engines, or natural language interfaces. By training the model on a dedicated dataset, the system avoids generic or overly broad matches, improving the relevance of the output. The ranking step ensures that users or downstream systems receive a prioritized list of potential matches, optimizing efficiency and accuracy in semantic processing tasks.
20. The computer program product of claim 18 wherein the grouping is preceded by encoding sentence embeddings contained in the corpus.
21. The computer program product of claim 20 where the encoding sentence embeddings comprises at least one process selected from the group consisting of a Language-Agnostic Bidirectional Encoder Representations from Transformers Sentence Encoding (LABSE) process, and a Robustly Optimized Bidirectional Encoder Representations from Transformers Pretraining Approach (RoBERTa) process.
22. The computer program product of claim 20 wherein the encoding sentence embedding is followed by, prior to the grouping, performing dimensionality reduction on the encoded sentence embeddings.
23. The computer program product of claim 22 wherein the dimensionality reduction comprises performing at least one process selected from the group consisting of a Uniform Manifold Approximation and Projection (UMAP) process, and a t-Distributed Stochastic Neighbor Embedding (t-SNE) process.
24. The computer program product of claim 18 wherein the grouping comprises performing at least one process selected from the group consisting of a Kmeans clustering process, a Concensus clustering process, and a cluster splitting process.
25. The computer program product of claim 18 wherein the creating of the set of candidate intent names comprises performing label generation.
26. The computer program product of claim 25 wherein the label generation comprises performing at least one process selected from the group consisting of a Generative Pre-trained Transformer 2 (GPT-2) process, a Bidirectional Encoder Representations from Transformers (BERT) process, and a simplification process.
27. The computer program product of claim 18 wherein the rating comprises ranking for most likely to least likely using a statistical model trained using a dataset for semantic similarity matching of labels to full sentences.
28. The system of claim 19 wherein the grouping is preceded by encoding sentence embeddings contained in the corpus.
The system relates to natural language processing (NLP) and information retrieval, specifically addressing the challenge of efficiently organizing and retrieving information from large text corpora. The core problem involves grouping related sentences or documents in a way that preserves semantic meaning while optimizing computational efficiency. Traditional methods often struggle with scalability and accuracy when processing large datasets. The system first encodes sentence embeddings from the corpus, converting each sentence into a dense vector representation that captures its semantic meaning. This encoding step uses advanced embedding techniques, such as transformer-based models, to ensure high-quality representations. The embeddings are then used to group related sentences or documents based on their semantic similarity. The grouping process may involve clustering algorithms, such as k-means or hierarchical clustering, to organize the embeddings into meaningful clusters. The system may also include preprocessing steps, such as tokenization or normalization, to improve the quality of the embeddings. Additionally, the system may support dynamic updates, allowing new sentences to be added to the corpus and integrated into existing groups without requiring a full reprocessing of the entire dataset. The overall approach enhances information retrieval by enabling faster and more accurate access to semantically related content.
29. The system of claim 28 where the encoding sentence embeddings comprises at least one process selected from the group consisting of a Language-Agnostic Bidirectional Encoder Representations from Transformers Sentence Encoding (LABSE) process, and a Robustly Optimized Bidirectional Encoder Representations from Transformers Pretraining Approach (RoBERTa) process.
30. The system of claim 28 wherein the encoding sentence embedding is followed by, prior to the grouping, performing dimensionality reduction on the encoded sentence embeddings.
31. The system of claim 30 wherein the dimensionality reduction comprises performing at least one process selected from the group consisting of a Uniform Manifold Approximation and Projection (UMAP) process, and a t-Distributed Stochastic Neighbor Embedding (t-SNE) process.
32. The system of claim 19 wherein the grouping comprises performing at least one process selected from the group consisting of a Kmeans clustering process, a Concensus clustering process, and a cluster splitting process.
The invention relates to a system for organizing data into meaningful groups, addressing the challenge of efficiently categorizing large datasets into coherent clusters. The system employs advanced clustering techniques to improve the accuracy and reliability of data grouping. Specifically, the system performs at least one of three clustering processes: K-means clustering, Consensus clustering, or cluster splitting. K-means clustering partitions data into K distinct clusters based on distance metrics, optimizing cluster centers to minimize intra-cluster variance. Consensus clustering combines multiple clustering results to produce a more robust and stable grouping by aggregating consensus across different methods. Cluster splitting refines existing clusters by further dividing them into smaller, more homogeneous subsets when necessary. These processes enhance the system's ability to handle complex datasets, improving the precision of data categorization. The system is particularly useful in applications requiring high-dimensional data analysis, such as machine learning, bioinformatics, and customer segmentation, where accurate clustering is critical for downstream tasks. By integrating these clustering techniques, the system provides a flexible and adaptive approach to data organization, ensuring better performance in diverse analytical scenarios.
33. The system of claim 19 wherein the creating of the set of candidate intent names comprises performing label generation.
34. The system of claim 33 wherein the label generation comprises performing at least one process selected from the group consisting of a Generative Pre-trained Transformer 2 (GPT-2) process, a Bidirectional Encoder Representations from Transformers (BERT) process, and a simplification process.
This invention relates to a system for generating labels for data, particularly in the context of machine learning or natural language processing. The system addresses the challenge of creating meaningful and accurate labels for training datasets, which is critical for improving model performance. The system includes a label generation module that processes input data to produce labels using advanced natural language processing techniques. The label generation module employs at least one of several processes: a Generative Pre-trained Transformer 2 (GPT-2) process, which generates human-like text based on input prompts; a Bidirectional Encoder Representations from Transformers (BERT) process, which provides contextual understanding of words in a sentence; or a simplification process, which reduces complexity to enhance clarity. These processes enable the system to create labels that are either highly detailed, contextually aware, or simplified, depending on the requirements. The system may also include a data preprocessing module to prepare input data for labeling and a validation module to ensure the accuracy and consistency of the generated labels. By leveraging these techniques, the system improves the efficiency and effectiveness of data labeling, which is essential for training robust machine learning models.
35. The system of claim 19 wherein the rating comprises ranking for most likely to least likely using a statistical model trained using a dataset for semantic similarity matching of labels to full sentences.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
December 16, 2020
November 22, 2022
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.